ANLY482 AY2017-18T2 Group 11 Project Overview

HOME

PROJECT OVERVIEW

ANALYSIS & INSIGHTS

PROJECT MANAGEMENT

DOCUMENTATION

ANLY482 MAIN

Final		Initial

Motivation

The lack of use of a VRS to provide delivery routing solutions is one that resonates strongly with some SME companies operating within the logistics sector. A possible reason is due to the high cost associated with the purchase of such a software. To illustrate, Paragon – a VRS provider in the United Kingdom, revealed that a system designed for a Delivery Service Company with a fleet size of 100 vehicles can expect license fees of up to £50,000. To exacerbate this situation, the listed price has yet to include additional cost such as the training cost and the maintenance fee that is usually required with the use of the software. This example serves to exemplify the high cost associated with a VRS and it can be observed that this is not in line with the purchasing capabilities of a typical SME. Thus, the team believes that it is even more crucial to identify and create low-cost solutions so as to ensure SMEs, too, have vehicle routing tools at their fingertips.

In addition, another underlying motivation for this research is due to the lack of relevant skillset in the current job market. To be more specific, in order to derive a VRS, it is imperative of the developer to possess domain expertise in the geospatial and computer science field. However, the lack of low-cost solutions in the market seems to indicate a possibility that such skills are scarce. Besides, given the team dynamics, it seems that the team is well-poised to derive a VRS for the SMEs.

Project Objective & Goal

Like all business models, operating with such a business model is not without its flaws. This form of business model has resulted in the company being susceptible to the integrity and capabilities of external parties – Temporary Drivers. As such, even if the company were to know the number of parcels to deliver in the following day, it might still be unable to accurately determine the number of Temporary Drivers needed. While the company has pre-existing solutions in solving this issue, they are still interested in exploring alternatives that will help improve their capabilities in this aspect.

In response, the team suggests that the company can engage in a systematic method to guide its decision on deciding the number of Temporary Drivers to employ. This could be accomplished by utilising a “Logistics Application” – Vehicle Routing Software (VRS), which is able to provide delivery routing details such as the time and route taken when a predetermined number of drivers is hired to complete the days’ worth of delivery. Ultimately, when the company is armed with these details, the company would not be at the mercy of its external contractors.

Methodology

To provide operational recommendations from the given dataset, we will thoroughly examine the dataset via the following four-step approach:

1. Data Exploratory

As the dataset is provided in Excel format, little data preparation is required by the team. Following which, the team would use methods such as summary statistics, to determine if there are any inconsistencies, missing and invalid values in the dataset.

2. Data Cleaning

As errors such as outliers and invalid values could lead to inaccurate results, the data must be cleaned to ensure that it is suitable for further analysis. Based on the dataset, the two most probable data errors are inconsistency data and missing or invalid values.

3. Data Analysis

After cleaning up the relevant data, an in-depth analysis will be performed on the data to gain meaningful insights. Based on preliminary discussion, we will be looking into these 4 analytical methods in analysing the data:

a) Time series analysis – As the data provided contains time-series variables, the team will be performing Time Series Analysis on the data. This will allow the team to identify many trends such as those pertaining to the number of Drivers. This would then aid the team in forecasting the optimal number of drivers required for future deliveries.

b) Frequency distribution & Maximum likelihood – During the data cleaning process, the causes for invalid values will be recorded. Frequency distribution will be used to determine the frequency of each causes. After which, maximum likelihood will be performed to identify which reasons contributed most significantly to the invalid values. These factors will be analysed in greater depth as they are the primary reasons for a failed delivery attempt.

c) Cluster Analysis – By using variables such as delivery area, number of parcel and size of parcel, the team will be able to profile its consumer segments and analyse how it changes over time. With this information, the team can then better estimate the number of drivers required for each region. This will help the company to optimise the number of driver required for each area, and potentially reduce the operation costs.

d) Correlation & Regression – Correlation Analysis will be performed to identify the relationship between explanatory variables. Besides correlating the variables, the team will also attempt to perform regression on explanatory variables against outcome variables. This analysis will allow the team to determine the relationship between the variables and derive many conclusions pertaining to operational costs. For example, the team will be able to understand how variables such as quantity and weight of parcel can significantly impact operation cost.

4. Data Visualisations

Lastly, we will also be looking at creating a dashboard with the following visualisations which will ultimately help the team present its recommendation. Some of the visualisations that will be derived are:

a) Spider Chart – To visualise the reason that contribute to data inconsistency

b) Time Series Line Chart – To forecast the optimal number of drivers needed in the future based on past data

c) Heatmap – To identify the number of drivers needed at the various location

Technology Used